Typicality based on soft aggregations in crisp object oriented databases

A method to automatically compute typical objects for a crisp object oriented database is proposed, in which typical objects are regarded as representatives of a fuzzy majority of the stored objects. A vector space model is adopted in which the object instances of a class are represented by vectors. The typical object of a class is derived as the "closest" vector to the fuzzy majority of the instance vectors. The use of typical objects is investigated to support the computation of default values in case of null values, to provide a means for database summarization and fuzzification.